AgentX and Deepgram Integration
Connect AgentX and Deepgram to automate ai workflows. Trigger events from one app, run actions in the other and use AI to make smart decisions. Integrate apps in 3 simple steps - no code required.
- 3-day trial
- Cancel anytime
- Up to 10 minutes quick set up
- GDPR
- Data Encryption
- 2FA
- Local Data Region
Learn about AgentX and Deepgram
Detailed info you need to know about AgentX and Deepgram integrations in WEXTL. Explore which modules apps support and how you connect them.
How to connect AgentX to Deepgram
Set up your AgentX and Deepgram integration in 3 simple steps[2]. No coding required. Up to 10 minutes setup.
STEP 1
Connect AgentX and Deepgram
Save credentials in WEXTL once. We store them encrypted and reuse the authorization across every workflow[3]. No coding or API skills required.
STEP 2
Build a workflow
Paste selected AgentX and Deepgram nodes onto the canvas[4], or describe what you want and WEXTL wires the flow together with AI Builder.
STEP 3
Map your data
Choose which fields need to sync between AgentX and Deepgram. Use filters or a JS custom code step in between. Everything is on the canvas.
How does the integration between AgentX and Deepgram work in WEXTL?
WEXTL exposes actions and triggers for AgentX and Deepgram that help you connect apps and automate repetitive tasks[1]. A trigger is an event that happens, and an action is a step that performs when the event has occurred.
Deepgram triggersAgentX actions
Deepgram modules
- Watch Callback Results Starts the workflow as soon as Deepgram sends a callback result. Paste the inbound URL of this trigger into the callback address of a transcription step or Analyze Text.
- Analyze Text Runs Deepgram text analysis on pasted text or a text file. Switch on Analyze Sentiment, Detect Topics, Detect Intents or Summarize for each insight, and set Callback URL to get the result in the background.
- Create a Key Generates an API key with a Comment, Roles and optional Tags. Set Expiration Date to make the key stop working on that day, and copy the Key from the result right away.
- Create a Self-Hosted Credential Makes a credential for pulling Deepgram self-hosted images from a registry. Use Scopes to say what it may pull, which defaults to Self-hosted products, and Comment to label it.
- Create a Temporary Token Issues an access token that works for speech and text calls only. Set Lifetime (seconds) anywhere in the 1 to 3600 range, or leave it empty for 30 seconds.
- Create an Agent Configuration Saves a voice agent configuration to a Deepgram project. Add Metadata entries such as a name to store with it.
- Create an Agent Variable Defines an agent variable with a DG_NAME style name. Use Value for a number, true or false, or JSON, which is stored as that type.
- Create an Invite Invites a person by Email to a Deepgram project with the Role of Member, Admin or Owner.
- Delete a Key Removes an API key that belongs to a project.
- Delete a Project Deletes a whole Deepgram project. This cannot be undone.
- Delete a Self-Hosted Credential Deletes a credential used to pull self-hosted images.
- Delete an Agent Configuration Deletes a saved voice agent configuration.
- Delete an Agent Variable Removes a variable used by voice agents.
- Delete an Invite Cancels an invitation sent to a project.
- Generate Flux Speech Turns Text into audio with a Flux Voice such as flux-alexis-en. Use Expressivity for a calmer or livelier delivery, and write {pause:500ms} markers into the text for pauses.
- Generate Speech Converts up to 2000 characters of Text into audio with an Aura Voice. Set Speed anywhere in 0.7 to 1.5, and pick an Audio Format such as MP3 or FLAC.
- Get a Balance Reads a single balance, giving its Units and Amount.
- Get a Billing Breakdown Breaks project spending down between Start Date and End Date. Narrow it by Key, Deployment, Tag or Line Item, and use Group By to split the report.
- Get a Key Returns the Member record linked to one API key.
- Get a Model Looks up one Deepgram model: its Name, Canonical Name, Architecture, Languages and Version, and whether it supports Batch or Streaming.
- Get a Project Fetches one Deepgram project with its Project ID, Name and MIP Opt Out setting.
- Get a Project Model Reads a model within one project. Output includes the Name, Version, UUID and Formatted Output.
- Get a Request Retrieves a past request by the Request ID of a Transcribe or Speech result, or of List Requests. Output includes the Path, Response, Code and Callback.
- Get a Self-Hosted Credential Reads one self-hosted credential, giving its Member and Distribution Credentials.
- Get a Usage Breakdown Splits project usage into periods between Start Date and End Date. Use Only Requests That Used and Only Requests That Did Not Use to include or leave out requests by feature.
- Get an Agent Configuration Reads one saved voice agent configuration with its Config, Metadata and timestamps.
- Get an Agent Variable Looks up an agent variable and shows its Key, Value, Created At and Updated At.
- Get Project Usage Summarizes project usage starting at Start Date and ending at End Date, which default to the project start and today. Filter by Endpoint, Method, Deployment, Key, Model or Tag.
- Leave a Project Removes you from a Deepgram project, and you lose access to it.
- List Agent Configurations Finds voice agent configurations in a project, with Agent ID, Config and Metadata. Use Limit to cap how many are returned.
- List Agent Think Models Shows the language models an agent can think with, giving ID, Name and Provider for each.
- List Agent Variables Lists the variables that voice agents of a project use, each with a Variable ID, Key, Value and timestamps.
- List Balances Finds project balances with the Balance ID, Amount and Purchase Order ID of each.
- List Billing Fields Lists the options for billing reports over a date range, with Accessors, Deployments, Tags and Line Items in the output. Set Start Date and End Date to bound the range.
- List Invites Shows invitations sent for a project with the Email and Scope of each.
- List Keys Shows each API key in a project along with its Member.
- List Member Scopes Shows the Scopes held by a single person on a Deepgram project.
- List Members Lists project members with Member ID, Email, First Name, Last Name and Scopes.
- List Models Lists Deepgram models, speech-to-text first and then voices. Models Deepgram no longer recommends are included only on request.
- List Project Models Lists the models that belong to one project. Use Limit to set how many are shown.
- List Projects Lists your Deepgram projects, giving the Project ID and Name of each.
- List Purchases Gathers purchases recorded in a project, with Order ID, Expiration, Amount, Units and Order Type.
- List Requests Searches past requests of a project within a date range, filtered by Status, Endpoint, Method, Deployment, Key or Request ID.
- List Self-Hosted Credentials Pulls up the credentials for self-hosted installs held by a project, each with a Member and Distribution Credentials.
- List Usage Fields Lists the filterable values in usage over a date range, including Tags, Models, Processing Methods and Features.
- Make an API Call Sends a free-form request to the Deepgram REST API, signed with your saved credential.
- Remove a Member Ends the membership of a person in a Deepgram project.
- Transcribe an Audio File Transcribes an uploaded Audio File with Deepgram speech to text. Choose a Model, set Language, and enable Smart Format, Paragraphs, Detect Speakers or Redact as needed. For recordings over about 25 MB use Transcribe Audio from a URL.
- Transcribe Audio from a URL Fetches a recording from a public Audio URL and transcribes it. Deepgram downloads it itself, and audio over 10 minutes can time out, so set Callback URL for long files.
- Update a Member Scope Changes what a project member may do by applying a Role in place of the old one.
AgentX modules
- Create a Conversation Opens a new conversation on one of your AgentX agents. Set Type to Chat, Search or Ecommerce to decide how the agent will handle it.
- Create Speech Turns Text into spoken audio through the AgentX voice API. Set Language, Gender, Speed or Accent to shape the voice, and turn on Robot for a mechanical tone.
- Create Speech Stream Synthesizes Text as a streamed audio response from the voice API, so playback can begin before the whole clip exists. Use Language, Gender, Speed or Accent to adjust the voice.
- Delete a Conversation Context Clears the memory context stored for a conversation held with an agent in AgentX.
- Get a Conversation Reads one conversation of an agent together with its messages. Create a Conversation or List Conversations gives you the conversation to read.
- Get a Message Trace Looks up the prompt trace linked to a message by its Message ID. Pass the result to Get a Trace for the full detail.
- Get a Trace Looks up the detailed prompt trace for a Trace ID, the record of how an agent built its answer. Get a Message Trace finds the ID for a given message.
- Get an Agent Fetches the details and current configuration of a single agent in your AgentX account.
- Get Your Profile Reads the profile of the user who owns the API key this credential uses, which shows which AgentX account a credential belongs to.
- List Agents Finds every agent your API key is allowed to use in AgentX, each coming back as a separate result.
- List Conversations Collects the conversations held with one agent, ready to read or continue in later steps.
- Make an API Call Makes an authorized free-form request to the AgentX REST API with your saved credential.
- Send a Message Posts a Message into an existing conversation and returns the Status and Response of the agent. Use Agent Mode to switch between Chat and Search, and Context to set how much history counts, with 0 meaning as much as possible.
- Send a Message and Wait Sends a Message and holds until the agent finishes the whole reply, buffered up to 10 MiB or 180 seconds. Use Context to cap how much earlier conversation counts.
- Send a Message and Wait for JSON Sends a Message, optionally with Attachments or Audio, and waits until the agent completes a JSON reply. Set Context to restrict the history the agent draws on.
- Send a Message with an Attachment Uploads a File into a conversation for the agent to read. Add a Message to go along with the file, and use Context to control how much earlier chat is considered.
- Send Speech Recognition Submits an audio Sample (Base64) for transcription under a Call ID and Seq ID. AgentX posts the recognized text to your Callback URL, and you can leave Language empty for auto detection.
- Update a Conversation Context Seeds a conversation with earlier user and assistant Messages as memory, without triggering a chat turn.
- Update a Message Vote Sets the Vote on a message, found by its Message ID. Use Agent and Trace ID to tie the vote to the exchange it rates.
Explore alternative integrations
There are other apps[9] supported by WEXTL that can be paired with AgentX and Deepgram integrations. Pick any from the list below if the current combination isn't the one you need.
Frequently asked questions
Common questions about connecting AgentX and Deepgram.
References
List of sources used to gather information about the AgentX and Deepgram integration.
- AgentX and Deepgram integration source code, module specs and credentials implementation.
- Our AgentX and Deepgram integration documentation, structured input and output parameters, scopes, and authentication methods.
- Official vendor API documentation for AgentX and Deepgram, used to verify field meanings, authentication flows, and anything our own specs don't fully capture.
- WEXTL® visual workflow builder, anonymized data from our technical support team.
- AgentX integration information
- Deepgram integration information
- WEXTL® Scheduling
- WEXTL® Pricing and Trial
- WEXTL® Apps Library
How we write our apps documentation
Explore how WEXTL® crafted AgentX and Deepgram integration page documentation. Used sources, the role of AI, and how the team validates the result.

